The Constitutive Proteome of Human Aqueous Humor and Race Specific Alterations

Aqueous humor (AH) is the fluid in the anterior and posterior chambers of the eye that contains proteins regulating ocular homeostasis. Analysis of aqueous humor proteome is challenging, mainly due to low sample volume and protein concentration. In this study, by utilizing state of the art technology, we performed Liquid-Chromatography Mass spectrometry (LC-MS/MS) analysis of 88 aqueous humor samples from subjects undergoing cataract surgery. A total of 2263 unique proteins were identified, which were sub-divided into four categories that were based on their detection in the number of samples: High (n = 152), Medium (n = 91), Low (n = 128), and Rare (n = 1892). A total of 243 proteins detected in at least 50% of the samples were considered as the constitutive proteome of human aqueous humor. The biological processes and pathways enriched in the AH proteins mainly include vesicle mediated transport, acute phase response signaling, LXR/RXR activation, complement system, and secretion. The enriched molecular functions are endopeptidase activity, and various binding functions, such as protein binding, lipid binding, and ion binding. Additionally, this study provides a novel insight into race specific differences in the AH proteome. A total of six proteins were upregulated, and five proteins were downregulated in African American subjects as compared to Caucasians.


Introduction
Aqueous humor (AH) is the fluid in the anterior and posterior chambers of the eye. It is produced by the non-pigmented ciliary body epithelium primarily through active transport of ions and solutes into the posterior chamber [1][2][3][4]. From the posterior chamber, the AH enters the anterior chamber via the lens and iris. After supporting the metabolic requirements of the avascular tissues of the anterior segment, the AH mainly exits the eye via the trabecular meshwork/Schlemm's canal into the episcleral veins, known as conventional outflow. AH outflow also occurs via an alternative route through the ciliary muscle bundles into the supraciliary and suprachoroidal spaces, which is known as uveoscleral outflow [5].
AH is an integral component in many ocular health functions, including nutrient and oxygen supply, the removal of metabolic waste, ocular immunity, and ocular shape and refraction [6][7][8]. The dynamics of AH and the fine balance between production and drainage is essential in maintaining the physiological intraocular pressure (IOP) [2].

Aqueous Humor Sample Preparation
The aim of this study was to characterize all of the proteins present in the human aqueous humor and we did not utilize immunodepletion to remove abundant proteins. Aqueous humor samples (60 µL) were lyophilized and subsequently reconstituted in 30 µL of 8 M urea in 50 mM Tris-HCl (pH 8). 20 mM Dithiothreitol (DTT) was then added to the mixture in order to reduce cysteine residues, followed by alkylation with 55 mM iodoacetamide. 240 µL of 50 mM ammonium bicarbonate buffer was added in order to reduce urea concentration to below 1 M. Total protein concentration was measured while using a Bradford assay kit (Pierce, Rockford, IL, USA), according to the manufacturer's instructions. The digestion of proteins was performed using a 1:20 ratio (w/w) of Trypsin (Pierce, Rockford, IL, USA) at 37 • C overnight. Figure 1 shows a schematic of the workflow involved in the AH sample preparation and proteomic quantification.  Figure 1. Liquid chromatography/Mass Spectrometry (LC-MS/MS) workflow for proteomic analysis of human aqueous humor. Samples were digested using trypsin and were analyzed using an Orbitrap Fusion Tribrid mass spectrometer coupled with an Ultimate 3000 nano-UPLC system. Proteins were identified and quantified using Proteome Discoverer (ver 1.4; Thermo Scientific, Waltham, MA, USA) followed by statistical analysis using the R Project for statistical computing (https://www.r-project.org/).

LC-MS/MS Analysis
The trypsin-digested samples were analyzed using an Orbitrap Fusion Tribrid mass spectrometer coupled with an Ultimate 3000 nano-UPLC system in order to perform in-depth proteomic characterization. Reconstituted peptides (6 µL) were trapped and washed on a Pepmap100 C18 trap at the rate of 20 µL/min using a gradient of 2% acetonitrile in water with 0.1% formic acid for 10 min. Subsequently, the peptide mixture was separated on a Pepmap100 RSCLC C18 column using a gradient of 2% to 40% acetonitrile with 0.1% formic acid for 120 min at a flow rate of 300 nL/min. Eluted peptides from the column were introduced into the Mass Spectrometer via nano-electrospray ionization source (temperature: 275 • C; spray voltage: 2000 V) and analyzed via data-dependent acquisition in positive mode. Orbitrap MS analyzer was used for precursor scan at 120,000 FWHM from 300 to 1500 m/z. MS/MS scans were taken while using an ion-trap MS analyzer in top speed mode (2-s cycle time) with dynamic exclusion settings (repeat count 1, repeat duration 15 s, and exclusion duration 30 s). Collision-induced dissociation (CID) was used as a fragmentation method with 30% normalized collision energy. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE [28] partner repository with the dataset identifier PXD022463.

Protein Identification and Quantification
For the protein identification and quantification, raw MS data were processed using Proteome Discoverer software (ver 1.4; Thermo Scientific, Waltham, MA, USA) and then submitted for SequestHT search against the reviewed-manually annotated Uniprot-SwissProt human database with 20,385 entries. The following search parameters were used: 10 ppm precursor mass tolerance and 0.6 Da product ion tolerance; static carbidomethylation (+57.021 Da) for cysteine, dynamic oxidation (+15.995 Da) for methionine, and dynamic phosphorylation (+79.966 Da) for serine, threonine, and tyrosine. Proteins that contain similar peptides, which cannot be differentiated based on MS/MS analysis alone, were grouped in order to satisfy the principles of parsimony. A report comprising the identities and spectrum counts (number of peptide-spectrum match) for each protein was then exported as a semi-quantitative measure for relative protein levels that were detected in the AH sample. Figure 2 shows an example of LC-MS/MS analysis of one AH sample.
for 10 min. Subsequently, the peptide mixture was separated on a Pepmap100 RSCLC C18 column using a gradient of 2% to 40% acetonitrile with 0.1% formic acid for 120 min at a flow rate of 300 nL/min. Eluted peptides from the column were introduced into the Mass Spectrometer via nanoelectrospray ionization source (temperature: 275 °C; spray voltage: 2000 V) and analyzed via datadependent acquisition in positive mode. Orbitrap MS analyzer was used for precursor scan at 120,000 FWHM from 300 to 1500 m/z. MS/MS scans were taken while using an ion-trap MS analyzer in top speed mode (2-s cycle time) with dynamic exclusion settings (repeat count 1, repeat duration 15 s, and exclusion duration 30 s). Collision-induced dissociation (CID) was used as a fragmentation method with 30% normalized collision energy. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE [28] partner repository with the dataset identifier PXD022463.

Protein Identification and Quantification
For the protein identification and quantification, raw MS data were processed using Proteome Discoverer software (ver 1.4; Thermo Scientific, Waltham, MA, USA) and then submitted for SequestHT search against the reviewed-manually annotated Uniprot-SwissProt human database with 20,385 entries. The following search parameters were used: 10 ppm precursor mass tolerance and 0.6 Da product ion tolerance; static carbidomethylation (+57.021 Da) for cysteine, dynamic oxidation (+15.995 Da) for methionine, and dynamic phosphorylation (+79.966 Da) for serine, threonine, and tyrosine. Proteins that contain similar peptides, which cannot be differentiated based on MS/MS analysis alone, were grouped in order to satisfy the principles of parsimony. A report comprising the identities and spectrum counts (number of peptide-spectrum match) for each protein was then exported as a semi-quantitative measure for relative protein levels that were detected in the AH sample. Figure 2 shows an example of LC-MS/MS analysis of one AH sample.

Statistical Analysis
The peptide-spectrum match (PSM) values from LC-MS/MS analysis were quantile normalized, and then log2 transformed to achieve normal distribution. For each protein, the detection rate (proportion of samples in which the protein was detected) was quantified. The proteins that were detected in a majority of samples (>50%) were examined in detail to see whether certain protein families were enriched in human AH. These commonly expressed proteins were also associated with gene ontology terms, including biological processes, cellular components, and molecular functions, using the "goana" function from "limma" (ver.3.40.6) R package. Adjusting for confounding variables, including age, sex and hypertension, differential expression analyses were performed using negative binomial regression, in order to discover differences in protein levels between African American and Caucasian subjects. The p-values were adjusted for multiple testing using the FDR method. All of the statistical analyses were performed using the R Project for Statistical Computing (version 3.5.1).

Protein Content of the Human Aqueous Humor
A total of 2263 unique proteins were identified in 88 aqueous humor samples (Table S1). These proteins were divided into four categories that were based on their detection in the number of samples: High (n = 152; detected in >75% of samples), Medium (n = 91; detected in 50-75% of samples), Low (n = 128; detected in 25-50% of samples), and Rare (n = 1892, detected in <25% of samples) ( Figure 3A). Figure 3B shows the sample-to-sample variation in the levels of these proteins (the coefficient of variation). The majority of proteins in the "High" group show low sample-to-sample variation, indicating the uniformity of expression across samples. As the mean expression decreases, the coefficient of variation increases from high to rare proteins. Table 2 shows a complete list of 152 proteins found in at least 75% of AH samples.  Figure  3A). Figure 3B shows the sample-to-sample variation in the levels of these proteins (the coefficient of variation). The majority of proteins in the "High" group show low sample-to-sample variation, indicating the uniformity of expression across samples. As the mean expression decreases, the coefficient of variation increases from high to rare proteins. Table 2 shows a complete list of 152 proteins found in at least 75% of AH samples.

Gene Ontology Enrichment Analysis
A total of 243 proteins that were detected in at least 50% of the samples were considered as the constitutive proteome of human aqueous humor. Gene ontology enrichment analysis was performed in order to discover the biological processes, cellular components, and molecular functions associated with the constitutive proteome ( Figure 4). The top enriched categories among the biological processes include organonitrogen metabolic process (136 proteins), protein metabolic process (127 proteins), transport

Network and Pathway Analysis
Ingenuity Pathway Analysis (IPA) was used to discover the protein-protein interaction networks in the constitutive proteome (243 proteins) of human aqueous humor. Figure 5 presents the three topscoring networks. Several members of the Apolipoprotein, Complement, and SERPIN families were part of the top-scoring network ( Figure 5A). The second-highest scoring network consisted of 56 proteins, which are involved in tissue development, protein synthesis, and cellular compromise ( Figure 5B). The third network includes several members of the Immunoglobulin and IGF families and other proteins that are involved in protein synthesis, humoral immune, and inflammatory

Network and Pathway Analysis
Ingenuity Pathway Analysis (IPA) was used to discover the protein-protein interaction networks in the constitutive proteome (243 proteins) of human aqueous humor. Figure 5 presents the three top-scoring networks. Several members of the Apolipoprotein, Complement, and SERPIN families were part of the top-scoring network ( Figure 5A). The second-highest scoring network consisted of 56 proteins, which are involved in tissue development, protein synthesis, and cellular compromise ( Figure 5B). The third network includes several members of the Immunoglobulin and IGF families and other proteins that are involved in protein synthesis, humoral immune, and inflammatory responses ( Figure 5C). IPA analysis also revealed that 21 canonical pathways were significantly enriched among the constitutive proteins observed in the AH ( Table 4). The highly enriched canonical pathways include acute phase response signaling (40 proteins), LXR/RXR activation (33 proteins), FXR/RXR activation (32 proteins), clathrin-mediated endocytosis signaling (18 proteins), complement system (17 proteins), and coagulation system (14 proteins).

Aqueous Humor Proteins Associated with Race
Analyses were performed in order to discover race-specific differences in the AH proteome (differentially expressed in African Americans as compared to Caucasian subjects). A total of six proteins were upregulated and 5 proteins were downregulated in African American subjects (Table 5) Figure 6 shows the boxplots depicting the distribution of these proteins in the African American and Caucasian subjects.

Discussion
This study provides the proteomic repertoire of human AH while using a larger sample set and highly sensitive mass spectrometry technology. The low abundant proteins have higher variation and poor reproducibility due to random nature of detection of proteins in mass spectrometry analysis. This study provides a reference AH proteome, which can be used in order to enhance the interpretation of results in future studies. We identified 243 proteins in at least 50% of samples, which we refer to as the constitutive proteome of human aqueous humor.
A comparison of our study with a previously published study by Chowdhury et. al. [2] revealed significant overlap in the proteins identified in human AH. We detected more than 79% of the 355 AH proteins that were identified in the previous study using nano-LC-ESI-MS/MS. Also, in the previous study, the samples were divided into three matched groups and 206 proteins were found in all three groups. A comparison of these 206 proteins with the constitutive AH proteome of our study (243 proteins) revealed >70% overlap [2].
Our comprehensive proteomic analysis revealed that five protein families are highly enriched in human aqueous humor, including apolipoproteins, complement proteins, immunoglobulins, IGF family proteins, and serine protease inhibitors (SERPINs). Apolipoproteins are proteins that bind and transport lipids in biological fluids. Seven apolipoproteins, including APOA1, APOA2, APOA4, APOC3, APOD, APOE, and APOH, were highly abundant, whereas five apolipoproteins APOB, APOC1, APOLD1, APOF, and APOL1 were detected in less than 25% of samples. Consistent to our findings, APOA1, APOA2, APOA4, APOD, APOE, and APOH were also identified in previous studies [2,29,30]. Several members of this family were part of the top scoring protein interaction network identified while using IPA analysis.
The anterior chamber is immune privileged and relies on AH to maintain a pathogen-free environment. Our analysis identified 25 complement proteins from both the classical and alternative pathways. Eleven complement proteins, including CFI, C4B, C6, C8A, and C9, were detected in more

Discussion
This study provides the proteomic repertoire of human AH while using a larger sample set and highly sensitive mass spectrometry technology. The low abundant proteins have higher variation and poor reproducibility due to random nature of detection of proteins in mass spectrometry analysis. This study provides a reference AH proteome, which can be used in order to enhance the interpretation of results in future studies. We identified 243 proteins in at least 50% of samples, which we refer to as the constitutive proteome of human aqueous humor.
A comparison of our study with a previously published study by Chowdhury et. al. [2] revealed significant overlap in the proteins identified in human AH. We detected more than 79% of the 355 AH proteins that were identified in the previous study using nano-LC-ESI-MS/MS. Also, in the previous study, the samples were divided into three matched groups and 206 proteins were found in all three groups. A comparison of these 206 proteins with the constitutive AH proteome of our study (243 proteins) revealed >70% overlap [2].
Our comprehensive proteomic analysis revealed that five protein families are highly enriched in human aqueous humor, including apolipoproteins, complement proteins, immunoglobulins, IGF family proteins, and serine protease inhibitors (SERPINs). Apolipoproteins are proteins that bind and transport lipids in biological fluids. Seven apolipoproteins, including APOA1, APOA2, APOA4, APOC3, APOD, APOE, and APOH, were highly abundant, whereas five apolipoproteins APOB, APOC1, APOLD1, APOF, and APOL1 were detected in less than 25% of samples. Consistent to our findings, APOA1, APOA2, APOA4, APOD, APOE, and APOH were also identified in previous studies [2,29,30]. Several members of this family were part of the top scoring protein interaction network identified while using IPA analysis.
The anterior chamber is immune privileged and relies on AH to maintain a pathogen-free environment. Our analysis identified 25 complement proteins from both the classical and alternative pathways. Eleven complement proteins, including CFI, C4B, C6, C8A, and C9, were detected in more than 75% of the samples. Similar to the blood plasma, several members of the Immunoglobulin family of proteins were also identified in the AH. Immunoglobulins are involved in cell communication, defense response, and the regulation of metabolic processes. The presence of a wide array of immunoglobulins has been reported in previous studies indicating their existence in the AH of cataract and glaucoma patients [12,31,32].
Insulin-like growth factors and their binding proteins have been shown to play an important role in ocular functions. Ten IGF family proteins were identified in our analyses. Several IGFBPs in vitreous and aqueous humor have been previously reported [33]. However, the predominant serum carrier protein, IGFBP3, was present in less than 50% of AH samples, whereas IGFBP7 and IGFBP6 were highly abundant, indicating quantitative differences between the two fluids. IGFBP7 has been linked to hypertensive retinopathy and familial retinal macroaneurysms, indicating its role in retinal vascular pathology [34,35]. Furthermore, IGFBP7 was elevated in the AH of exudative age-related macular degenerative patients and is considered to be an anti-angiogenic agent in these patients [36].
Interestingly, after adjusting for confounding variables, including age, sex, and hypertension, we found 11 proteins that were differentially expressed between African American and Caucasian subjects, indicating race-specific differences in the AH proteome. Overall, six proteins were significantly upregulated, while five proteins were downregulated in African American subjects. Proteins related to immune responses such as IGKV1D-33, C6, C8A were present at elevated levels in African American subjects. Among the downregulated proteins, TSPAN-14 was at least two-fold lower in African Americans. A member of this family, TSPAN-12, was discovered as a therapeutic target for retinal vascular diseases, such as age-related macular degeneration and diabetic retinopathy [50]. Three other vision-related proteins, including RBP4, TTR, and PTGDS, were present in lower levels in the African American population. RBP4, a retinol transporter protein, is known to be involved in congenital eye disease [51]. TTR is a transport protein, which carries retinol-binding protein and is essential for the maintenance of photoreceptors, visual cycle, and perception [52]. PTGDS is a secretory retinoid transporter that is involved in the maintenance of the blood-retinal barrier. The difference in the vision-related proteins might be one of the contributing factors for increased risk of eye-related ailments in the African American population.

Conclusions
In conclusion, this study characterized the human aqueous humor proteome using the latest technology and a larger sample set. A total of 243 proteins, which were detected in at least half of the samples, were considered to be the constitutive proteome of human aqueous humor. Five protein families were highly enriched in the human aqueous humor proteome. Eleven proteins were significantly altered between African American and Caucasian subjects, indicating race-specific differences. The highly abundant aqueous humor proteins are involved in immune-mediated responses, transport, metabolism, and binding. The reliable characterization of the aqueous humor proteome will provide new insights into the factors that govern anterior segment homeostasis and aid in biomarker discovery in various eye disorders.
Supplementary Materials: The following are available online at http://www.mdpi.com/2227-7382/8/4/34/s1, Table S1: Complete list of proteins detected in 88 aqueous humor samples. Funding: This study is supported by the National Institutes of Health, National Eye Institute (Bethesda, MD, USA) grant# R01 EY029728 awarded to Ashok Sharma and grant# P30 EY031631 Center Core Grant for Vision Research.

Conflicts of Interest:
The authors declare no conflict of interest.